In our class, we had discussed the mechanism of the second price or the Vickrey auction and how the dominant strategy for each bidder was to bid the true value that the he has for the object. One feature about the second price auctions we talked about was that there was only one round of bidding. Secondly, the objects that were being auctioned were non-perishable. Thirdly, we knew the “number” of objects that were being sold during the auction. So you must be wondering why exactly am I talking about these seemingly unimportant points in these auctions! The reason is that these are not always the cases in real world. Take for example, the case of search engines like Yahoo and Google allocating positions to advertisers on their pages.

Since the search engines can only show a limited number of ads each time a user enters a keyword, they need to sell these positions to the advertisers and something like second price auctions seems to be the right thing to do. These search engines use mechanisms based GSP (Generalized second price) auctions to sell these ad positions as opposed to the Vickrey-Clarke-Groves (VCG) mechanism. In the simplest GSP auction, for each keyword, advertisers submit the prices they are willing to pay. The bidder with the maximum value gets the position on the top and he pays the price of the bidder who got the position below him. Thus, a bidder which won the position i pays the cost bid by the bidder of the position i+1 . Even though the multiple positions available generalize second price auction, unlike VCG, bidding the true price is not the dominant strategy in GSP.

The main factors that set bidding using VCG apart from bidding using GSP are that while bidding for keywords, you are actually playing infinite rounds of bids. And thus you could change your bids based on the bids of other bidders in the previous rounds. Also, the search engines are effectively selling perishable items because if they don’t have an ad for a particular period of time, they are actually losing money. And not just that, even after bidding, since the policy is pay per click, the bidders don’t know exactly how many(much) things till after the bidding has been done and some number of users have clicked on their ads. Using all these factors, we can prove how bidding the true price does not remain the dominant strategy in this business. For an in-depth proof, please read

http://www.benedelman.org/publications/gsp-060801.pdf

As learned in class, eBay functions as a second-price sealed-bid auction where bidders submit bids, and the “winner” pays the second highest bidder’s value. In such an auction, the bidder’s dominant strategy is to bid his or her true value—how much the item is worth to him or her. The underlying logic is that if the bid deviates from the bidder’s value—either higher or lower—the bidder will pay more than the object is worth (to him or her) or irrationally lose the auction, respectively. Using this logic, the bidder seemingly should bid his value early on and let the auction pan out thereafter. However, many eBay practices, particularly effective multiple bidding and late bidding, defy second-price logic. Sharad Borle, Peter Boatwright, and Joseph Kadane of Statistical Science explain how the seemingly counterintuitive tactics are often successful, effectively winning the auction below the bidders’ values.

To confirm the high incidence of multiple bidding and late bidding, data were from August 2001 to February 2002 across 27 products in 15 consumer product categories. The variables of the study were those relevant to multiple bidding and late bidding, namely the level of individual participants (i.e. experience) and individual’s bid timing. Level of experience was determined using eBay’s quantified feedback system. Ultimately, the authors discovered that experienced bidders were more likely to be active in bidding towards the end. In juxtaposing the final winning price (achieved by a late bidder) against the market value of a good, they concluded that a vast majority of “late winners” realized great deals. The studies also proved that experienced bidders did not engage more heavily in multiple bidding, but rather were more likely to bid once. In fact, the study proved such statistically significant results that the authors concluded that eBay as a whole deviates from regular second-price sealed-bid auctions solely because of these two additional strategic bidding behaviors.

The study merely claimed that there was a high incidence of late bidding and multiple bidding on eBay but failed to explain such behaviors. I would posit that eBay hosts such a variety of bidders with varying degrees of experience that bids are often not in line with the object’s value. Furthermore, the majority of eBay’s auctions span multiple days; thus, the nature of the auctions themselves differs from standard second-price sealed-bid auctions in that bidders are not constantly bidding. As a result, as many bidders’ likely have busy schedules, late bidding accommodates that. Finally, multiple bidding is probably a result of bidders’ changing perceived value of the object throughout the auction.

Borle, Sharad. “The Timing of Bid Placement and Extent of Multiple Bidding: An Empirical Investigation Using eBay Online Auctions.” Statistical Science 21.2(2006) 194-205. 05 MAR 2008 <http://projecteuclid.org/DPubS/Repository/1.0/Disseminate?handle=euclid.ss/1154979821&view=body&content-type=pdfview_1>.

The Game theory, which we learned in class, can be applied to athletes with doping. The game played is similar to prisoner’s dilemma. Say there are two athletes, A and B. A thinks if B doesn’t take any drugs, then it will be in A’s best interest to take them. A taking the drugs will give A performance edge against B; A has a better chance of winning.  On the other hand, if B takes drugs, it’s also in A’s best interest to take them. In that way B won’t have an edge over A. This results in A taking the drugs to be the dominant strategy.  The same can be applied to athlete B. So, if the game were to be played out, the Nash Equilibrium is (drugs, drugs).  If the two athletes trusted each other, then there would be no doping problems. This would lead to much better payoff, parallel to prisoner’s dilemma, where if the prisoners trusted each other, then the jail time would be much smaller for both prisoners. However, that is not the case in reality today. As we have learned in class, the Nash Equilibrium does not always give parieto optimal, nor maximum social welfare. Applying these terms to athlete’s dilemma, this pair of strategy (drugs, drugs) does not benefit either athletes, nor does it benefit the general population, who may be fans of such athletes. http://www.wired.com/politics/security/commentary/securitymatters/2006/08/71566BruceSchneier,Drugs: Sports’ Prisoner’s Dilemma 

No, it is not just my productivity secret. If you haven’t stopped already, perhaps this website might sway you:

http://www.coolwormvirus.info/w32-blaster-virus-w32-worm/w32-blaster-virus-w32-worm.html

I admit, I was apprehensive at first about clicking on a link that had the word “virus” in it three times. But I figured, I’m in a public Cornell computer lab, what’s the worse that can happen (to me)? (Hackers crash the Cornell network, student records are compromised… in hindsight maybe it wasn’t such a great idea. This could come back to bite me later) Nevertheless, I am happy to report that the website is clean and your records are safe.

We live in an age where information travels at the speed of light, and all business relies on efficient communication. It is only a matter of time before everyone on the face of the earth and your neighbor’s dog will all be connected to the internet. We’ve become so dependent on this electronic networking medium to the point that I’ve seen people in the same room communicate with each other via instant messenger (IM) rather than by mouth. Remember when people use to write letters? How about when news had to travel via the messenger boy on horseback?

We live in a time when not only people can get sick, but our computers can come down with the sniffles as well. The easiest way for your computer to get sick? IM. According to John Sakoda of IMLogic (now owned by Symantec Anti-Virus):

“IM viruses and worms are growing exponentially….Virus writers are now shifting the focus of their attack to instant messaging, which is seen as a largely unprotected channel into the enterprise.”

As if attacking your computer wasn’t bad enough, hackers can easily steal your confidential information. And not only are you not safe, neither are your family and friends. In a large network, having individual nodes compromised rarely poses a threat to the integrity of the entire network. However, in this case, it is possible for one node to unknowingly infect many other nodes, which can potentially give rise to a cascading effect. Over a billion people use the internet, with new users joining the network each day. Take for instance your 80 year old grandmother, my mom who just learned how to forward email last week, and my friend’s 7 year old kid brother. None of these people have bad intentions intentions when they log onto the web, but each of them can easily be the victims of other people’s bad intent.

As discussed in Tipping points, an epidemic requires three key elements to sustain itself. First of all, there are connectors, the hackers that are well-connected in the network and are in a position to target many people. This also applies to the victims of hackers who are well-connected socially, as they have the potential to cause the most damage. Secondly, there are the mavens, the people who thrive in acquiring knowledge and sharing it with others. Unfortunately, malicious individuals of this type also exist. According to the article, “hacker manuals” are readily found on the internet that can guide you in gaining access to any system. Also, the article claims that “hacker-friendly providers” exploit their position over customers by providing software that facilitates access to your computer. Finally, there are the salesmen, those unique individuals that are highly skilled at their craft. In the same way that a skilled salesman can change your feelings and preferences, a skilled hacker can easily manipulate your computer. Perhaps in the context of internet security, these three factors may not necessarily sum up to an epidemic, but it should certainly raise awareness of your online habits, especially on IM.

At least there’s still Google chat.

The average company attains the services of a shipping company to move products with hours of telephone calls and patience. The Mars corporation does it a little different though, they instead use the Internet and game theory. Mars has devised an online auction house, Freight Traders, that brings shippers and carriers together. Tenders, or oligations to perform, that used to take months now run in weeks. One shipper claims that freight costs have been reduced by 50%. According to game-theory rules, Mars has changed the game to provide win-win logistical solutions for everyone involved.

Shipping costs in the transportion sector can account from anywhere netween 3% and 6% of the budget for most companies. Logistical planning for these massive corporations is not an easy task and there exists a certainty that shippers and carries will eventually find themselves locked in a price haggling battle. The shippers and carriers are essentially locked in a “zero-sum” game. In these situations, in terms of shipping costs, somebody is usually either going to win or lose completely with a happy medium rarely occurring.

Freight Traders is the first step in interfering with the zero-sum structure in the logistics world. It provides an auction-based broad internet community that excels at providing strategically priced outside options for shippers or carriers who initially would have been subjected to the mercy of one of their few choices. The second step is where game theory becomes involved by assisting Freight Traders in deciding when an auction should cease. 95% of the bids from carriers occurs in the last 2 hours of an auction and Game Theorists suggest that “Sniping,” or withholding a bid to the last possible second, hinders competition. To combat the issue Freight Traders has contrived a “soft ending” mechanism, which extends the deadline of the auction if someone bids in the last hour, ultimately removing the allurement of sniping .

Savings for companies that utilize Freight Traders are estimated to be in the order of 5% - 8%. These savings are generated mainly through efficient bartering and proper truck allocation. The number of freightless trucks in logistics plans, otherwise known as “Empty Running,” has decreased by 20% since the inception of online auction-based trades.

http://www.fastcompany.com/magazine/69/smartcompany.html?page=0%2C0

Hezbollah and the Prisoner’s Dilemma

Can game theory solve the Israel-Lebanon war?

By Tim Harford

http://www.slate.com/id/2146867/

This article discusses how game theory relates to the Israel-Lebanon war.  If Hezbollah were to attack Israel, Israel will respond with a far greater counter attack.  Unfortunately, Israel’s strategy has not deterred Hezbollah’s attacks.

This article addresses how economic theories such as the prisoner’s dilemma can be overly simplistic.  The article describes a solution to the prisoner’s dilemma, differently than the Nash Equilibrium where both prisoners will confess:

“The champion was “tit for tat,” which begins by cooperating with its

fellow prisoner (staying silent) but punishes a squealer by confessing

on the next turn. Axelrod argued that “tit for tat” was successful

because it was easy to interpret, hard to exploit, began

cooperatively, and quickly forgave transgressions by returning to

cooperation. It has proved a magical myth: that you should speak

softly and carry a big stick, that “an eye for an eye” can produce

cooperation in unpromising situations. Axelrod’s idea was repeated in

a horde of popular science books.”

Here the article contradicts our homework solutions, where we found that:

“in the second play of the game, the players are essentially playing a

one-shot prisoner’s dilemma. In this game, C is a dominant strategy

for each player. So (C,C) will be played and they should both realize

that this is how the game will be played the second time round. This

happens regardless of how the first game was played. As a result, in the first game the players should realize that nothing they do can affect the outcome of the second play of the game. This means that the first play of the game is also essentially a one-shot prisoner’s dilemma and again we would expect to observe (C,C). This logic suggests that we would expect to see (C,C) played repeatedly.”

Overall I am not convinced that the prisoner’s dilemma is a useful analogy to the Israel-Lebanon war.  There are far more issues, options, and players in the situation.

www.pandora.com

http://www.bostonherald.com/business/technology/general/view.bg?articleid=1075879&srvc=rss

The listener’s access to music has been a form of great debate in the last decade. With Napster, the illegal downloading of music spread quickly and the media industry was dealt a hard blow. In the aftermath, many other programs arose and mimicked or expanded upon the new gateway to music. At Cornell, many students use Direct Connect to get their media files. This system allows people within the Cornell network to share a collection of their media library. Direct Connect, and programs similar to it, have pushed the boundaries of legally downloading music. An alternative to these questionable programs comes in the form of Pandora. Pandora is, in essence, an online radio but it does so much more than dispense preplanned music. When first arriving at the Pandora website, one is prompted to search for a song or artist that interest her. Pandora matches the search, and then has the user verify the correct song or band (there are many songs and bands with similar names). Once the details are correct, Pandora’s search system connects your music choice with a plethora of other songs that share similar attributes to the original selection. When listening to one of the selected songs, there are options to agree or disagree with its placement. This interaction with the system allows it to be updated through user feedback. The specialization that Pandora offers differentiates it from the radio and from other online music websites, and allows the user to come across new music that will interest her. There are two downsides to Pandora: it is only accessible online, and there is a limit to the number of songs the user can skip through. The people behind Pandora are working towards making it more accessible (making it “mobile”), so that problem could be fixed relatively soon. The other problem, the inability to skip over songs one dislikes, is a matter of legal complexities. Still, on the radio there is no ability to skip over songs so the ability to do so, although it is constricted, provides a better experience for the user. Pandora has a novel idea for dispensing music to the user and if the creators are able to make their product mobile and develop a broader user bas then they will definitely be a key part of music access in the future.

The foundation and spread of Pandora can be viewed in many different network settings. The primary network setting that comes to mind when thinking about Pandora is the technology network that assists the play list compilation. The nodes in this network are songs and the edges are the common attributes attached to the songs. These attributes are determined through the user interaction and the foundation set in place by the creators, which helps segue into the second network that Pandora has created: the social network. Users are themselves nodes, with common music interests connecting them acting as the edges. There is the opportunity to friend people who have these similar music interests, creating a network between users. In addition, since the users can give feedback to Pandora there is also a diverse network of users and Pandora employees together manipulating the song network. Pandora has also created links along with the songs that the user is currently listening to. This link (edge) connects the Pandora website (node) with either an iTunes or Amazon website (nodes). The inclusion of this accessory allows users to have direct access to buying the song that they are listening to. The Pandora website has created and expanded many networks within the music community. Doubtlessly, as Pandora grows so will its networks, creating benefits for both the users and the artists who are getting exposure.

Myrimis (my-risk-managenment+information-system) is a product of Nanomatic, a company which utilizes and initiates novel types of nano-information technology. Myrimis, “a secure place on earth,” [i] operates on Nanomatic’s Nanonavi, “personal navigation on earth.” [ii] Myrimis, in conjunction with Nanonavi, provides social-networking software that represents an evolutionary step in the company’s development of complex data networks and geographical information systems.

Social networks are an integral part of everyone’s life—a fact upon which social-networking sites, like Nanomatic’s Myrimis, aim to capitalize. As we’ve discussed in class, social networks consist of a number of people (represented in graphs as nodes) connected by a series of ties (represented in graphs as edges). The graph models we’ve studied in class display distance only in terms of the shortest number of ties (edges) between two people (nodes). This means that two people who are best friends (two nodes connected directly by a single edge on a graph) could, in real life, be thousands of miles apart. Social-networking software such as Myrimis provides a platform whereby users can more easily connect to everyone in their social network, regardless of physical distance.

According to Myrimis’ nanowiki, Myrimis is a “location-based services platform that provides social network functionality combined with location-based media and telematics.” [iii] Akin to other social-networking sites, such as Myspace or Facebook, Myrimis offers a range of services to allow users to generate their own set of social ties on Myrimis.

Unlike other social-networking sites, however, Myrimis also employs GPS technology (Nanonavi). Using GPS, Myrimis “enables users to modularly configure and develop their “own world,” consisting of family, friends, pets, possessions such as cars, houses, and boats, and personal data including photos, videos, and holiday diaries.” [iv]

Myrimis provides an intriguing way to observe and keep track of social networks. Theoretically (provided that the proper GPS equipment is available at each respective location), Myrimis makes it possible to keep track of any given person on your social network. You can group people by their distance or relationship to you—family, relatives, friends, coworkers, buddies from an activity such as golf or yoga, etc.—and perform a number of functions thereon out. You can “geo-tag” (with specific locations) photos or videos, conveniently view points of interest (POIs) displayed for your specific area, and search for people or places in any location.

Myrimis’ more advanced services provide ways to actively manage your Myrimis network. For instance, Myrimis can notify you if your dog has left your yard, alert you if someone has tripped your home security alarm, and allow you to monitor the functions (lights, thermostat, security system) of your home or vacation house while you’re away. With Myrimis, nearly everyone and everything in your social network is just one click away.

Myrimis succeeds in bridging some of the physical distance between people within social networks by bringing them closer in other ways (internet, phone, etc.). Two best friends, thousands of miles away, can now visually observe each other’s activities and whereabouts at any point in time; and, subsequently, they can engage in real-time interaction. In that moment of real-time interaction, those two best friends are no farther apart than the nodes and edges by which they are depicted in a graph (well, figuratively speaking). In effect, by spanning the physical distance between people, Myrimis allows the real world to more closely resemble the theoretical graphs in which it is represented.

[i] http://www.myrimis.com

[ii] http://www.nanomatic.net/wiki/index.php/Nanonavi

[iii] http://www.nanomatic.net/wiki/index.php/Myrimis

[iv] http://www.nanomatic.net/index.php?option=com_content

[all photos are screen-captures from http://www.myrimis.com]

Lately in class we’ve been discussing social exchange theory, specifically, as it applies to the one-exchange game. As it turns out, some of its premises are largely applicable to the exchange dynamics of intimate relationships – interactions that take place in the marketplace for sex and love if you will. People look to maximize profit (or personal gain) in romantic exchanges (consistent with the ideas of rationality and self-interest raised in class), and stable outcomes in terms of who dates whom are largely contingent on each person’s options within the network (that is, the potential romantic transactions that can be made with neighboring nodes).

A paper by Rijt and Macy (2006) raises the idea that the lower your dependence on a relationship, the greater your bargaining power within that relationship. Perhaps, if you’re a person with many desirable qualities (sexual attractiveness, wealth etc) and many potential romantic partners you are likely to gain relatively more from a romantic exchange than someone who is highly dependent on the romantic attentions of only a few people. In class, we described social exchange as a bargaining procedure in which one’s power is determined by one’s relative dependence on others in the network. Rijt and Macy contend that intimacy is also inherently economic; sex and love are forms of social exchange. What makes someone dependent on someone else for sex/love/romance? According to Rijt and Macy, the exchange occurring in the existing relationship must be strictly better than the most attractive alternative. This seems to be related to the concept of stable outcomes. That is, if the sum of the two individuals’ best outside options exceeds the total surplus that can be achieved from trading with each other then there will be no deal (or no romantic/sexual interaction as the case is here). Taking this one step further, the paper proposes an essential ingredient for relationship longevity: greater payoff in the current relationship than one could plausibly expect from potential alternative partners. Long-lasting relationships are similar to “stable” social exchange.

In a similar paper, Baumeister and Voh (2004) raise the idea of the “principle of least interest”. A person is relatively more powerful on account of desiring an exchange less than the other person. For example, in a relationship, the person who is relatively less in love is more powerful. Might the person who is relatively more in love (and consequently less powerful) be more willing to settle for an egregiously unequal split – much like those who occupy the weakest positions in the one-exchange game? Being very much in love presumably means that a person is far less amenable to his outside options, or perhaps, does not perceive that he has them…

Baumeister and Voh also take up the argument that sex (in heterosexual relationship) is like a resource possessed by females and “purchased” by men through such “payment” as companionship, promises of marriage, fancy dinners etc. If we believe this model (and there is some evidence to suggest that men want sex more than women) then sexual exchanges are not unlike the bipartide market interpretation of social exchange. Here, women are the sellers and men are the buyers… Lastly, Baumeister and Voh’s paper raised and explored some interesting questions. Might advances in birth control that have come about in the last 50 or so years have decreased the cost of sexual exchange for women, increasing their payoff for a given exchange, but weakening their bargaining power? The paper explores this possibility. Might men need to bring less to the negotiations as a result? Along similar lines, has the market value of female sexuality declined due to social and cultural changes such as increased economic opportunities for women and increased sexual permissiveness? It’s interesting to think about how “power” might be influenced by more than just network position.

Rijt and Macy (2006), “Power and Dependence in Intimate Exchange” - http://muse.jhu.edu/login?uri=/journals/social_forces/v084/84.3van_de_rijt.html (If you have access)

Baumeister and Voh (2004), “Sexual Economics: Sex as Female Resource for Social Exchange in Heterosexual Interactions” -

http://search.bwh.harvard.edu/concourse/900/articles/BaumeisterSexEcon.pdf

 http://www.nytimes.com/2007/07/26/health/26fat.html?scp=7&sq=social+network&st=nyt Last summer, an article entitled “Find Yourself Packing It On? Blame Friends” appeared in the New York Times. The story details a study in the New England Journal of Medicine suggesting that when a person gains weight, their close friends also tend to gain weight. The study followed a group of participants from 1973 to 2003, tracking weight gain and social connections in a sample from a small town. The online link to the article also includes a video component, showing the growth of the network over time. It is evident from this video that individual’s nodes turn to green (the color denoted for those who are technically obese) shortly after the nodes that they share edges with turn green.  The study found that friends are more likely than neighbors or relatives to mimic the weight gain of individuals. The article reports that, when one close mutual friend became obese, “the other had a 117 percent increased chance of becoming obese, too.” This would seem to make sense if one considers the conditions under which people gain weight. Most people do not dramatically gain weight overnight, but rather as a function of their day-to-day activities and lifestyles. Thus, it would make sense that friends, the people who we choose to spend our time with, would have a larger influence on these day to day activities than neighbors and family members, who we see only on occasion. For instance, though we may eat a particularly large thanksgiving dinner with our relatives, or join our neighbors for an occasional walk around the neighborhood, it is our friends who we share meals with day in or day out or engage with in regular exercise. Because of these habits, it would make sense to qualify our friends as “strong ties,” and our neighbors and relatives as “weak ties,” because of the involvement these different groups of people have in our food and exercise routines. It makes sense, then, that our friends’ weight gain, compared to that of other groups, has the largest effect on our own weight gain.  The New York Times article infers that the study may suggest that the “obesity epidemic” is a result of a few people beginning to gain weight from something in the environment. In his book The Tipping Point, Malcolm Gladwell suggests that epidemics and messages spread through a network because of “connectors,” people that, for whatever reason maintain more social ties than other people and function as links that bring people together. It is interesting to consider what type of people might be deemed connectors in the spread of an epidemic such as obesity. It seems that people who are highly respected in their social networks might serve this function. The Times article suggests that when one person gains weight, people’s ideas of body image shift and they think it is acceptable to gain weight, also. It makes sense that people who are highly respected and emulated in other weighs would have more of an effect on people’s shifting images of obesity. Specifically, if people who take care of their bodies in all other ways start to gain weight, others may see this as the norm and begin to gain weight as well. Though it sounds superficial, if a person looks pleasing in all other ways, their weight gain may not be noticeable, and this may encourage people to emulate their appearance, even if it means that they begin to gain weight themselves.